Economic Voting and Incumbent Mayoral Elections in Canada: 1997-2010

Cameron D. Anderson Assistant Professor

Department of Political Science The University of Western Ontario

London, CA cander54@uwo.ca

Josh Morgan PhD Student

Recruitment and Development Officer Local Government Program

Department of Political Science The University of Western Ontario

London, CA jmorga8@uwo.ca

Paper prepared for presentation at the Annual Meetings of the Canadian Political Science Association, Waterloo, CA, May 2011.

2

Introduction Voting is a core right of citizens in a democracy and often the only political act for most of the

electorate. Given the centrality of voting to empower legitimate democratic authority, the objective of elucidating why citizens vote the way they do has resulted in a large sub-field of political science that is concerned with assessing the impulses of voters in democracies the world over. While this collective interest has fostered a great many studies of voting behaviour in federal and provincial elections in Canada, there remains an evident gap in our comparable understanding of electoral behaviour at the local government level in Canada. As Sancton recently noted: “the harsh reality is that we know very little about the municipal voting behaviour in Canada” (2011, 193). While we do not endeavour to fill this entire omission, this paper represents a sustained effort to explore how one factor- economic conditions - influences local-level electoral behaviour in Canada. In the analyses which follow, we test the key proposition underlying economic voting theory at the local level in Canada. Briefly, the theory of economic voting suggests that incumbent office holders are more (less) likely to be re-elected when the economy is doing well (poorly). Using Census Metropolitan Area unemployment data from Statistics Canada data and incumbent mayoral election results, we assess the extent to which incumbent mayoral vote support is a function of local economic conditions. To our understanding, this is the first such effort to explore the relationship of economic conditions and mayoral elections in Canada.

In the sections which follow we provide a brief overview of the model of economic voting and a discussion of how the model might be applied to the evaluation of incumbents and candidates for mayoral office in Canada. Following a detailed elaboration of our data, modeling choices and results, we explore the implication of our findings and conclusions for understanding municipal voting in Canada. Economic Voting and Canada

The act of casting a ballot is often the central and only means of active participation in democracy for many Canadian voters. The goal of explaining how and why individuals vote the way they do has engendered a significant sub-field of political science committed to understanding the motivations of voters in Canada and around the world. Major approaches to the study of voting behaviour began in the United States in the 1940’s. The Columbia School emphasized the importance of long held and relatively stable group memberships and identities such as class, religion, ethnicity and gender as a key factors shaping voter preferences (Lazarfeld et al. 1944). While having empirical applicability, this approach was centrally criticized for not being able to capture variation over time in voter preference by scholars who ultimately became known as the Michigan School (Campbell et al. 1960). The Michigan approach emphasized socio-psychological constructs like partisanship (an affective identification with a political party) as well as issue and leader evaluations all of which contributed to the formation of voter preferences. Since the 1960’s, these theoretical models of voting behaviour have been adapted and tested within the Canadian context (e.g. Clarke et al. 1979; Mesiel 1973; Nevitte et al. 2000).

The theoretical approach of economic voting clearly falls under the ‘Michigan School’. The fundamental premise of economic voting is that when voters enter the ballot box a prime consideration influencing their vote decision is an evaluation of incumbent performance. If the incumbent (candidate, leader, party or government) has performed well according to some standard then the chances of support and re-election are good. By contrast, if voters deem the incumbent to have performed poorly on issues of importance then the incumbent may well be less likely to keep office. One such basis of evaluating incumbent performance is perceived management of the economy.

The logic underlying “economic voting” is that, in the formation of evaluations of incumbent performance and ultimately vote decisions, citizens base these judgments, in part, on economic conditions. If the economy (either real or perceived) is doing well, then the incumbent is more likely to

3

be rewarded with re-election. Alternatively, if the economy is performing poorly, the chance of electoral defeat for the incumbent becomes much greater. Stated simply, the central theoretical proposition of economic voting model is that economic conditions influence election outcomes: an incumbent is rewarded/punished at the ballot box for good/poor economic conditions.

A key presumption underlying the approach is that governments (at whatever level) have some influence over economic conditions. Commenting on the relationship between government policy and economic outcomes, Norpoth writes:

Governments have good reason to believe that an economy in good health will earn them the support of the electorate. They enact policies that, in their judgment, will promote that economic health and thus assure their hold on political power. A great deal of government activity is consumed with the shaping and adjusting of economic policy. After all, government itself is a giant among economic actors, with close to four-tenths of the gross domestic product going through its hands (1992, 51).1

While referencing the national government in Britain, the quote from Norpoth makes a strong case for the connection of economic performance and the actions of government in the management of it. This is not to suggest that the state of the economy is only a result of government action (at any level). Clearly, the state of the economy at any point in time is a function of past economic management (perhaps from different previous administrations) not to mention exogenous international influences (such as foreign investment, international trade relations and global market fluctuations). Nonetheless, governments make decisions that respond to these outside influences and these can additionally be a reflection of competence that form the basis of evaluation.

Of course, when talking about the “economy” central distinctions ought to be addressed. One clear division within the literature is that between the ‘real’ and ‘subjective’ economy. The real economy pertains to actual objective economic conditions like unemployment rates, inflation or GDP growth. All else being equal, approval and re-election of incumbents are more likely when the actual economy is performing well (for instance, low or declining unemployment, low and stable inflation and/or positive GDP growth) (e.g. Lewis-Beck and Stegmaier 2000). In contrast to the real economy is the ‘subjective’ economy. The subjective economy is made up of perceptions of economic conditions. These perceptions may or may not be correct or accurate but they most clearly exist in the minds of individual voters. In the context of economic voting, these subjective perceptions can be part of what drives the vote decision: if voters think or believe that the economy is good/bad then they are more/less likely to vote for the incumbent candidate, party or executive. This dichotomy of conceptualizing the economy is characteristic of the economic voting literature (see Lewis-Beck and Stegmaier’s review of economic voting literature (2000)). Additional distinctions regarding the subjective economy are made within the literature that, while extending well beyond the parameters of this paper, may be helpful to outline. Briefly, a differentiation is made between the collective or sociotropic economy (or a city, province or country) and a household or personal economy. The literature tends to find that sociotropic economic evaluations have a greater influence on vote decisions than perceptions of personal economic situations. A second distinction is based on the temporal period. This contrast is between evaluations of past economic performance (retrospective) and future economic performance (prospective). On this basis, voters may evaluate governments on the basis of past economic conditions and their management of them and/or differentiate amongst competing options for whom they think will best manage the economy in the future. The literature tends to find that retrospective economic evaluations have the strongest influence on vote choice (Lewis-Beck and Paldam 2000).

1 The four-tenths figure comes from the case of Britain. In other countries, the extent of economic involvement of

government (at all levels) may be larger or smaller than this figure.

4

In the Canadian case, there is a significant body of literature which observes the effects of economic conditions on vote choices and election outcomes. Over the past twenty-five years, a range of studies have observed effects of objective economic conditions (and subjective perceptions of them) on federal incumbent support (Anderson 2008; Clarke and Kornberg 1992; Gelineau and Belanger 2005; Happy 1986, 1989, 1992; Nadeau and Blais 1993, 1995).2 Other studies have considered how federal economic voting differs by region (Godbout and Belanger 2002). At the provincial level, past work in Canada has observed evident effects of economic conditions (both objective and subjective) on provincial incumbent support (Anderson 2008, 2009; Gelineau and Belanger 2005; Tellier 2006).

Given this record of consistent observation of economic effects at both the federal and provincial levels in Canada, this paper poses the core question as to whether these kinds of effects might also be observed in voting at the municipal level – in particular, for the office of mayor. Local Economies and Local Government Responsibility In looking to apply and test the core insights of the economic voting model to mayoral elections in Canada, we first begin with a discussion of how and why voters in Canadian municipal elections might look to the economy as a means of evaluating mayoral performance before considering the small body of literature which has considered municipal elections in Canada. In general, municipalities lack the strong fiscal tools of the federal and provincial governments to influence and shape macroeconomic policy. Federal and provincial governments have significant authority to set taxation rates (income tax, sales taxes, trade tariffs etc.) as well as the ability to borrow (when needed) for economic stimulus or basic program spending. Given the giant economic status that the federal and provincial governments have in Canada relative to municipalities, why should or could economic conditions have an influence on election outcomes at the local government or mayoral level? A first response might be that there are some formal economic management tools available to mayors and their municipal councils. For instance, local governments do have land-use planning and zoning powers which can influence the ongoing viability of existing business as well as the attractiveness of one municipality over another for new business. Additionally, local governments in Canada have control over commercial property taxes, as well as water and sewer rates. In the interest of attracting business to locate in their municipality, municipal governments may look to leverage their control over these kinds of rates to improve the likelihood of investment being located within their borders. Arguably, any one of these tools may be utilized by a local government to make their municipal jurisdiction as attractive to investment and job creation as possible.

Of course, tools such as tax rates and land-use planning which are available to municipalities are relatively blunt and disparate instruments. Indeed, based on these characteristics, it may be difficult for municipal governments to harness these means of economic maneuverability in a particular direction or for a specific purpose. Additionally, use of these kinds of tools for economic prosperity tend to create economic impacts over a longer term time frame than the usual municipal electoral cycle of three or four years. As a result, it may be very difficult for the actions and decisions of mayor to take the desired effect on the economic health of the local community the current term in office. While these sorts of tools available to mayors and local governments may have some objective effect on local economic conditions, what may provide a strong foundation for an influence of local economic conditions on mayoral elections is that voters hold the perception that mayors have definitive authority over the local economy.

2 While Carmichael (1990) finds no evidence for economic effects on federal government support, Nadeau and

Blais (1993) suggest that this is due to different methodological choices in the modeling of aggregate economic effects.

5

There may be a number of reasons underlying the perception (and contrary to objective scope of local authority) among voters that local governments and mayors have an important influence over the local economy. The first reason for the possible misperception of local authority deals with political information levels of voters. In general, there is widespread evidence that the electorate’s level of information about politics is quite low (Delli Carpini and Keeter 1996; Fournier 2002). As a result of low levels of information about politics, there may well be inaccuracies in voters’ views of the competencies and scope of authority of local government to influence local economic conditions. Indeed, looking at voters’ understanding of the federal division of powers within Canada, Cutler observes widespread confusion about the nature of authority between the federal and provincial governments in such areas as health care, the economy, taxes and energy (2008). Applying these insights to an accurate knowledge of authority at the local level, it is highly plausible that voters, in general, lack a clear understanding of the limited scope of municipal and mayoral authority to intervene and change direction in the local economy.

A second and related reason why the perception of local government influence in the economy may persist is the public pronouncements and pledges of sitting mayors and candidates in mayoral elections. It may be that contributing to the creation of widespread misperceptions of local economic authority is the rhetoric conveying this impression utilized by mayoral candidates. For example, recently defeated three-term incumbent mayor of London, Ontario, Anne-Marie Decicco-Best, attributed her mayoral election loss in October 2010, in part, to poor economic conditions. Quoted in the local London newspaper after her defeat, she stated: “I’ve served with a lot of other great leaders in other communities (who) I know unfortunately were not successful and the economy may have played a part in that for the entire region” (Maloney 2010). Similarly, the mayoral candidate who defeated DeCicco-Best, former federal Liberal MP Joe Fontana, campaigned on a promise to create 10,000 jobs in London over a five year period (Miner 2010). While London is just one example, a quick web search of Mayoral candidate web-pages from the 2010 municipal elections across the country shows that economic growth and job creation is a common plank for incumbents as well as challengers in mayoral elections. Taken together, this kind of anecdotal evidence suggests that mayors and mayoral candidates may themselves contribute to raising the expectations of the electorate for the ability of local government to be a prominent player in the economy.

Ultimately, however, the true capacity of a municipality to control the economic conditions within its borders does not matter (from an electoral perspective) if there is a perception by the electorate that they should be able to create jobs and positively control the local economy. If citizens believe local government and local mayors should be able to impact the local economy, and they hold them to account for the results of the local economic conditions, then the actual capacity of local governments to do so becomes a moot point. Our goal is to measure whether or not citizens do, in fact, use local economic indicators, like unemployment rates, as an economic voting tool for judging the performance of incumbent mayors. Economic Voting and Municipal Elections in Canada In contrast to the rich variety of work assessing the relationship of economic conditions and electoral outcomes at the federal and provincial levels in Canada, there are precious few which do the same at the local level.3 Indeed, Sancton notes: “the harsh reality is that we know very little about the municipal voting behaviour in Canada” (2011, 193). There have, however, been attempts to better

3 Outside of Canada, a few studies exist which consider economic influences on local election. Both Vermeir and

Heyndels (2006) and Bosch and Sole-Olle (2007) assess the role of tax policy and inter-municipal comparison (yardstick competition) in municipal election outcomes in Flanders and Spain, respectively.

6

understand the dynamics of municipal voting behaviour that have shed some light economic conditions and elections at the local level. The key study for our purposes is that of Cutler and Matthews (2005) who consider voting behaviour and the role of economic perceptions in the 2002 City of Vancouver election using individual-level survey data.4 Among their findings, Cutler and Matthews observe that voters who held a positive view of Vancouver’s economy (when asked about their perceptions of it over the previous 12 months) were 17 points more likely (than those who thought the economy was performing poorly) to support the incumbent party (2005).5 As these authors note, regardless of whether the local government has any control over local economic conditions, “voters appear to hold municipal governments accountable in a crude way for the nature of the times in the locale” (Cutler and Matthews 2005). Despite the fact that only about 5% of respondents indicated that they thought that the municipal government was most responsible for the local economy, Cutler and Matthews observe that economic perceptions still had a “powerful influence on the vote” (2005).

Given the preceding discussion of economic voting, the nature and perception of mayoral authority in Canada and the findings of Cutler and Matthews using subjective economic perceptions, our paper tests the following core proposition: better (worse) local economic conditions will positively (negatively) influence incumbent mayoral support in Canada. We consider this proposition using objective economic data in cases of incumbent mayoral elections in Census Metropolitan Areas from 1997 to 2010. The following section outlines in detail how we proceed to evaluate this central proposition. Data and Methods The economic and political data for the analyses in this paper come from two sources: Statistics Canada and city websites. In particular, the economic data come from Statistics Canada. From March of 1996, Statistics Canada has produced a monthly measure of unemployment in Canadian Census Metropolitan Areas (CMA’s).6 In sum, there are 31 CMA’s for which Statistics Canada has compiled this economic indicator.7 The beginning date of this series as well as the CMA designation provide the two greatest limitations on our data collection. In short, prior to 1996, there are no other available and widely consistent measures of economic conditions for major urban areas in the country. So, mayoral elections in CMA’s from 1996 onward provide the universe of cases to consider the question of economic effects on mayoral elections. Beyond the time period and CMA designation a number of additional factors influenced whether or not a CMA election was included in our data set. In the first instance, an election was included if there was an incumbent mayor running in the election. As the theory of economic voting suggests, it is only incumbents who can be rewarded or published for past economic conditions. So mayoral elections were only included that had an incumbent running for re-election. Second, the issue of municipal amalgamations posed numerous challenges to including many municipal elections for consideration. Our guiding principle in building the data set was to maximize the ‘fit’ between the geographic locale of economic conditions and the pool of electors voting for or against the incumbent mayor. On the one hand, we wanted to ensure that the CMA economic data pertained to the correct municipal jurisdiction.

4 There are other studies assessing electoral dynamics at the local level in Canada (notably Kushner et al. 1997 and

Stanwick 2000). While these studies make valuable contributions to understanding electoral dynamics at the local level in Ontario (Kushner et al. 1997) and Toronto (Stanwick 2000), neither explicitly consider the effect of economic conditions on electoral outcomes. 5 Vancouver is one a few cities in Canada that has political parties at the local level.

6 These data are found in CANSIM Table 2820116.

7 See appendix for list of CMA’s.

7

On the other hand, we wanted to ensure that the pool of electors who voted into office a mayor were also the same pool who had opportunity to re-elect or defeat that individual. For both of these reasons, amalgamations in Nova Scotia, Ontario and Quebec greatly reduced our ability to include mayoral elections in which the geographic fit of economic conditions and pool of electors was compromised. As a result of these limitations of data, we have 90 usable election events or cases for the period 1996 to 2011. These come from all provinces (less PEI – which in its entirety is a CMA). The data set is Ontario heavy in that almost half of the cases come from Ontario (47). The effects of economic conditions on mayoral elections are considered using one central dependent variable. This is a measure of electoral support for the incumbent mayor in each electoral event included in the dataset. This variable is calculated as the percentage of all votes cast which were cast for the incumbent mayor. Among the usable mayoral elections, the average incumbent support was 53.02%.

(Table 1 about here) As mentioned, we use the unemployment rate at the CMA level as our indicator of local economic conditions. Outside of being the only available economic indicator, CMA unemployment rate is an excellent measure of municipal economic conditions. Most aggregate analyses of economic voting use some form of unemployment rates as an indicator of economic conditions (Anderson 2009; Nannestad and Paldam 1994; Powell and Whitten 1993). Additionally, unemployment is a tangible economic condition in the lived experience of voters – friends and neighbours can lose their jobs (or be hired), large employers in the community may depart (or arrive) and various types of media tend to focus on unemployment rates in their economic coverage. For these reasons, unemployment at the CMA level is an excellent indicator of economic conditions (and the perception of those conditions) upon which the logic of economic voting is built. We have calculated the unemployment rate as the average of the monthly unemployment rates in the 12 months preceding the mayoral election. Constructing the measure of unemployment in this manner obviates the need to make assumptions regarding the time period prior to an election at which the unemployment rate might be more or most salient for a voter (is it 3 months prior to an election?; 6 months? etc.). We additionally calculated and estimated all models using a six month measure and substantive results were not different. The benefit of a 12 month calculation is that it smooths out the inevitable bumps in unemployment through a year that a 3 or 6 month measure might be more susceptible to. As a result, we believe that we are using a relatively conservative measure of unemployment rate. As listed in Table 1, the average CMA unemployment rate in the 12 months prior to a usable mayoral election was 6.89% The models control for other factors which might additionally influence the nature of electoral support for incumbent mayors. In particular, we control for the number of terms that an incumbent has previously served as mayor. In large measure, when conducting electoral analyses using aggregate data, the economic voting literature tends to find that the effect of a terms variable is negative: i.e. vote support slowly dissipates the longer a government is in office. Following these findings elsewhere, we would expect a similar effect. Of note, where an amalgamation has occurred we only count terms served for an incumbent mayor within the amalgamated city even if the incumbent served as a mayor for a previously existing constituent municipal entity. The second control variable we use is the number of candidates contesting a mayoral election. Arguably, the effect of this variable should be negative: the more candidates in the mayoral race the lower the overall support for the incumbent. A third control variable is previous support for incumbent mayor. Most aggregate analyses of economic voting include some sort of measure of political support which serves as a baseline of support for the incumbent mayor in next election. As a result our models include the percentage of popular vote that an incumbent mayor received in the previous mayoral election as an additional control.

8

Finally, because we consider mayoral elections from across the country, we include a regional dummy variable for the province with the largest number of cases: Ontario. We don’t have any necessary expectations about how this variable will perform other than to suggest that there may be some aspect of regional variation across the universe of cases in which elections within one geographic entity may be systematically more or less likely to produce greater incumbent support. In choosing Ontario for this purpose, we have no central theoretical rationale other than it is the locale of the greatest number of incumbent mayoral elections across our time period.

The effect of economic conditions on incumbent support is modeled using ordinary least squares (OLS) regression. This is an appropriate estimation strategy largely due to the nature of the dependent variable which is continuous. Results8

Our analyses assess the drivers of incumbent mayoral vote support when such individuals are seeking re-election. Results are presented in Table 2. We estimate the effects of economic conditions and control variables on incumbent support in the first model and then introduce the Ontario dummy variable and past vote in subsequent iterations – allowing us to more clearly isolate what effect, if any, there is of the unemployment rate. Results from model 1 suggest that a one unit increase in the 12-month average CMA unemployment rate prior to an election produces a 1.62 point decrease in electoral support for the incumbent mayor. In short, this result confirms the economic voting logic as applied to mayoral re-election – on average, as the unemployment rate increases incumbents in mayoral elections tend to lose support.

(Table 2 about here) Beyond the effect of unemployment on incumbent mayoral success, the coefficients for both

the number of terms in office and the number of candidates running in the election are negative. Specifically, for each additional term an incumbent mayor has served his or her vote support declines by 3.26 points. This result mirrors findings in other economic voting literature showing a negative effect on vote support of length in office. For each additional candidate in the mayoral election, incumbent mayors receive just under 1 point less of the popular vote. This result is expected as well. Note, however, that the constant for this model is 74.83. This value indicates the average level of incumbent support when all independent variables are set to zero. The sizable constant suggests that incumbents have a huge advantage over their rivals for winning office (perhaps driven by familiarity) but that with an increasing number of terms, this advantage slowly dissipates.

Model 2 introduces the dummy variable for Ontario. Of note, in model 2 the effects of unemployment, terms and number of candidates all remain negative (the expected direction). The

8 Before considering the results, some discussion of statistical significance in interpreting these results is

warranted. It is commonly accepted that when considering the effects of independent variables on dependent variables that the resulting p-values (generated from relative sizes of coefficients and standard errors) should be below, at the very least, 0.1. In analyzing data which constitute a sample from a larger population, the importance of a statistically significant result (indicated by a small p-value) is paramount. The ability of the researcher to infer any observed effect amongst the sample to a larger population is entirely dependent on the generation of a strong p-value (often at the 0.05 or 0.01 levels). This is because the researcher is not examining all available data points of a population but merely a sample (usually and hopefully random) of a population. The nature of the data used in this paper is different. Included in our data set is not only a sample of municipal elections but the entire population or universe – given the temporal and ‘fit’ limitations to creating the data set. As such, we are not concerned about extrapolating results observed here to a larger population – we are analyzing the population or universe of cases – in our situation, incidents of incumbent mayoral re-election in Canada since 1996. This is not to suggest that p-values do not matter in interpreting the strength of relationship but that they are not definitive in evaluating the substantive implications of results.

9

effect of the Ontario dummy variable is negative and quite large. Substantively, the coefficient of the Ontario variable suggests that relative to incumbent mayoral elections in all parts of the country other than Ontario, incumbent mayors in Ontario receive 7.78 points less of the popular vote controlling for the economy, terms and number of candidates. This is a substantive and large effect suggesting that incumbent mayors in Ontario face much stronger odds against re-election than those in the rest of the country.

The final model presented in Table 2 introduces the additional control variable for past vote.9 In the first instance, past incumbent vote share has surprisingly little effect on incumbent support both in coefficient size and statistical significance. This said, the effects of all other variables in the model remain in the same direction and both the terms and Ontario dummy variable reach some level of statistical significance.

In sum, although not reaching conventional levels of statistical significance, the results are suggestive of the theorized relationship of economic voting in mayoral re-election bids: as the economy worsens, incumbent mayors are punished for this poor economic performance. Additionally, both terms in office and number of candidates running in an election have suggestive effects in theorized (negative) direction on incumbent vote percentages. The coefficient for the Ontario variable was negative and quite large- suggesting that somewhat different dynamics may be underlying the election of Ontario mayors compared to the rest of the country. Finally, past incumbent vote has no effect on current incumbent vote share.

Based on the demonstrated effect of the Ontario dummy variable, it appeared to us that there may be something qualitatively different about the dynamic of these variables in municipal elections within Ontario as compared to outside Ontario. Based on this justification, we decided to re-estimate our core models separately for incumbent mayoral elections inside Ontario and outside of Ontario. These results are presented in Table 3.

(Table 3 about here) We first consider the results for Ontario. We estimate the core model including the unemployment rate, terms, number of candidates running and past vote. The coefficient direction of increases in CMA unemployment rates shifts from the earlier observed negative to positive – suggesting, contrary to economic voting wisdom, that as unemployment increases incumbent mayors in Ontario are rewarded. However, the result is statistically insignificant due to the large standard error. While the terms in office and number of candidates control variables are in the expected negative direction, neither exert a significant effect on incumbent support in Ontario. Finally, the coefficient for past vote registers a negative although insignificant effect. These results provide an intriguing baseline against which to consider the results for incumbent mayoral elections in rest of the country. Of the core model estimated in Ontario, results are highly insignificant and shed little light on the drivers of incumbent mayoral support within Ontario. For the cases of incumbent mayoral elections in the rest of the country, we estimate two models: a core model (with a regional dummy variable for Quebec) and then one with past vote added. The results of both stand in stark contrast to that observed for Ontario. We consider the core model first. Results from the core model indicate a significant and negative relationship between higher unemployment and incumbent vote share. For each one-unit increase in CMA unemployment rate, incumbent mayors receive about 4.3 points less of the popular vote. The result is highly significant (p<.01). Compared to the results for Ontario alone, it suggests that the economy has an important and large effect on incumbent mayoral elections outside Ontario but not within. Additionally, both number of candidates and terms in office have negative and significant effects on incumbent vote share. For

9 Due to difficulties in collecting data, we were unable to collect past vote for all incumbent mayoral elections.

Models which include past vote lose 8 elections from Ontario and 5 from outside Ontario.

10

each additional candidate running in a mayoral election outside Ontario, electoral support for incumbent mayors drops by 4.7 points. Similarly, for each additional term an incumbent mayor has served support, on average, is about 5.7 points lower. The Quebec dummy variable is small and statistically insignificant. The final model reported in Table 3 adds past vote to the estimation. In this most stringent test of the core economic voting proposition, previous results remain substantively similar and robust. Again, past incumbent support has no effect on current electoral strength. Beyond this, though, unemployment continues to have a significant and strong negative relationship with incumbent support- based on this model for each one percent increase in CMA unemployment, incumbent support drops by 3.07 points. Taken together, the results from these models suggest a strong role of economic conditions in shaping the electoral fortunes of incumbent mayors outside Ontario. 10 Given the contrasting economic effects on incumbent mayoral support between Ontario and the rest of the country, a central question is why might this evident discrepancy appear - why is it that the economy matters for incumbent mayoral support outside Ontario but not within? One response to this question and the confounding results observed centres on the basis upon which voters evaluate and perceive the economy. In particular, the ways in which these perceptions might systematically differ in Ontario as compared to the rest of the country. Previous work within the field of economic voting has considered the question of attribution of responsibility for economic conditions. Presumably to hold a government (or a mayor) accountable for the state of the economy (either positive or negative), a voter needs to assign at least some measure of responsibility for economic conditions to the government at hand. Previous work on the question of economic responsibility attributions in Canada has considered the ways in which voters make such determinations (Anderson 2006). In this work, Anderson found that voters in Ontario (relative to other provinces and regions) were much more likely to attribute responsibility for economic conditions to the federal government over that of the provincial government or other non-governmental actors (such as business, unions or the United States) (2006).

While this work didn’t specifically consider why Ontario voters were more likely to see economic responsibility in the hands of the federal government, a few reasons come to mind that largely deal with the subjective perceptions of voters. In the first instance, the federal government may be viewed as more prominent or responsible because the capital of the country is located in Ontario. It may also be the case in that the geographic position of Ontario within the country as a whole engenders a more centralist view of Ontario voters to both the economic and political core of the country. Finally, and perhaps as a result of both of these, there may be effects of political identity in economic responsibility attributions. We know that Ontarians are more likely than voters in the rest of the country to state Canada or Canadian as their primary political identity – particularly in comparison to the salience of provincial identities in other provinces such as Quebec or Alberta (Mendelsohn and Matthews 2010). Taken together, these points suggest a number of reasons for why voters in Ontario might tend to view the economy in Canada in more centralist terms than voters outside of Ontario.

Therefore, applying this discussion of economic perceptions and responsibility attributions of Ontarians versus voters in the rest of the country, we may have the beginnings of a plausible reason for the disjuncture in economic effects. This account suggests that voters outside of Ontario are more likely to view politics and economic conditions in more decentralist terms than Ontarians. Arguably, these decentralist views outside of Ontario may also shape views and perceptions of local levels of government and the economic responsibilities assigned to them. As a result, differences in perceptions

10

We should also note that the R2 value in the outside Ontario models is significantly greater than that for Ontario.

While some caution should be taken when using R2 values to evaluate the performance of models, the contrast

between the two models is notable and suggests a much better performing model for cases outside Ontario.

11

of the economy and economic responsibility may provide the context in which the stark contrast in economic effects on incumbent mayoral electoral support emerges between Ontario and the rest of the country. Conclusion This paper sought to test the proposition that economic voting takes place at the municipal level in Canada. Considering municipal elections across the country, results suggest that there is an evident, yet modest, expected relationship between economic conditions and incumbent mayoral support – worsening local economies weaken incumbent mayoral support. Our analysis then elaborated this analysis by considering the relationship in mayoral elections inside and outside of Ontario and the results were clearly sharpened – local economic conditions have a strong and significant influence on mayoral elections outside of Ontario but not inside. We interpreted this finding in light of how Ontarians tend to view the economy in more centralist terms than citizens in other parts of the country. We conclude by suggesting some implications of the findings of the paper. Clearly, with the exception of Ontario, local economic conditions have an evident influence on incumbent mayoral support. Regardless of the extent of actual policy levers that municipal governments and mayors might have to influence and shape the direction of local economies, these effects are likely driven by the perception that mayors and local governments have such ability. While this somewhat misguided perception may be a function of overall voter naivety, the paper’s central findings stand out as an important empirical observation regarding the role of the economy in Canadian municipal elections. In short, the economy matters.

Situated in the broader literature on voting behaviour, the findings of this paper may suggest the applicability of a performance or ‘valence’ model of municipal voting. First developed by Stokes (1963, 1992), valence issues are ones for which there is broad public agreement (either in support or opposition). At the local level in Canada, issues such as the economy or crime could be perceived as valence issues because all candidates for office are likely in agreement: economic health and public safety should be pursued. Accordingly, when issues such as these are important to an election, the electorate evaluates candidates for mayoral office on the basis of who they perceive is best able to govern on the issue (Clarke et al., 2009). As a result, the valence model of vote choice prioritizes past and/or anticipated performance on the issue in contrast to fundamental agreement or disagreement on the issue position itself. Because the economy is a valence issue, the findings of the paper suggest that this model of voting may be particularly applicable to understanding municipal and mayoral voting. Additionally, because the vast majority of cities in Canada lack political parties at the municipal level (which through partisanship can serve to structure vote choices), voters may rely more significantly on performance issues like the economy to reach their vote decision than they would in other electoral arenas such as provincial or federal elections.11

Based on these results, incumbents and candidates for mayoral office in Canada ignore the economy at the electoral peril. This is not to suggest that mayors have all the tools to be prime influential players in the economy as discussed above. However, when mayoral elections approach, incumbents and candidates alike would be wise to consciously consider the nature of their public pronouncements vis-à-vis the local economy. For the incumbent, if the economy is doing well, they should seek to highlight this strength and take as much credit as possible – building on the perception of voters that mayors have the ability to be influential actors in the local economy. When the local economy is not doing well, however, the incentive structure to use the economy as an election issue clearly changes. For the incumbent, they may want to downplay economic performance, highlight the

11

Of the municipalities under consideration in this analysis, only Vancouver and Montreal have political parties at the local level.

12

steps taken to maintain economic performance (it could have been worse if the incumbent did not do x, y and z) and perhaps blame the state of the local economy on decisions taken elsewhere at the provincial or federal levels (an age-old tactic of blame avoidance). For the non-incumbent mayoral candidate, highlighting the economy as a performance issue and one upon which they will be able to perform better manage may well be pivotal to using the local economy to their electoral advantage.

In sum, this paper makes a contribution to our understanding of the dynamics of municipal elections in Canada. The economy matters in incumbent mayoral re-election. This central finding provides a helpful window into the uncovering the decision calculus that voters make when casting ballots at the municipal level in Canada.

13

Works Cited Anderson, C. 2006. “Explaining Attributions of Responsibility for Economic Conditions in Multilevel

States: The Case of Canada.” Presented at the Annual meetings of the Canadian Political Science Association, Toronto, Canada, June 2006.

Anderson, C. 2008. “Economic Voting, Multilevel Governance and Information in Canada,” Canadian Journal of Political Science. 41, 2, June.

Anderson, C. 2009a. “Institutional Change, Economic Conditions and Confidence in Government: Evidence from Belgium,” Acta Politica. 44, 1, April.

Anderson, C. 2009b “Context and Sub-National Economic Voting in Canada.” Presented at the Annual Meetings of the American Political Science Association, Toronto, Ont., August, 2009.

Bosch, N. and A. Sole-Olle. 2007. “Yardstick competition and the political costs of raising taxes: An empirical analysis of Spanish municipalities.” International Tax and Public Finance 14: 71-92.

Campbell, A., P. Converse, W. Miller and D. Stokes. 1960. The American Voter. New York: Wiley. Carmichael, C. 1990. “Economic Conditions and the Popularity of the Incumbent Party in Canada.”

Canadian Journal of Political Science 34 (4): 713-760. Clarke, H., J. Jenson, L. LeDuc and J. Pammett. 1979. Political Choice in Canada. Toronto: McGraw-Hill

Ryerson. Clarke, H. D., A. Kornberg and T. J. Scotto. 2009. Making Political Choices: Canada and the United States.

Toronto: University of Toronto Press. Clarke, H. and A. Kornberg. 1992. “Support for the Canadian Federal Progressive Conservative Party

since 1988: The Impact of Economic Evaluations and Economic Issues.” Canadian Journal of Political Science 25: 29-53.

Cutler, F. 2008. “Whodunnit? Voters and Responsibility in Canadian Federalism.” Canadian Journal of Political Science 41, 3: 627-54.

Cutler, F. and J. S. Matthews. 2005. “The Challenge of Municipal Voting: Vancouver 2002.” Canadian Journal of Political Science 38, 2: 359-82.

Delli Carpini, M. and S. Keeter. 1996. What Americans Know About Politics and Why It Matters. New Haven: Yale University Press. Fournier, P. 2002. “The Uninformed Canadian Voter.” In Citizen Politics: Research and Theory in

Canadian Political Behaviour in ed. J. Everitt and B. O’Neill Don Mills, ON: Oxford University Press.

Gelineau, F. and E. Belanger. 2005. “Electoral Accountability in a Federal System: National and Provincial Economic Voting in Canada.” Publius-The Journal of Federalism 35: 407-24.

Godbout, J.-F. and E. Belanger. 2002. “La Dimension regionale du vote economique canadien aux elections federale de 1988 a 2000.” Canadian Journal of Political Science 35: 567-588.

Happy, J. 1986. “Voter Sensitivity to Economic Conditions: A Canadian-American Comparison.” Comparative Politics 10: 45-56.

Happy, J. 1989. “Economic Performance and Retrospective Voting in Canadian Federal Elections.” Canadian Journal of Political Science 22: 377-387.

Happy, J. 1992. “The Effects of Economic and Fiscal Performance on Incumbency Voting: The Canadian Case.” British Journal of Political Science 22: 117-130

Kushner, J. D. Siegel and H. Stanwick. 1997. “Ontario Municipal Elections: Voting Trends and Determinants of Electoral Success in a Canadian Province.” Canadian Journal of Political Science 30, 3: 539-53.

Lazarsfeld, P., B. Berelson and H. Gaudet. 1944. The People’s Choice. New York: Duell, Sloan and Pearce. Lewis-Beck, M. and Paldam, M. (2000) ‘Economic voting: an introduction’, Electoral Studies 19: 113–122. Lewis-Beck, M. and Stegmaier, M. (2000) ‘Economic determinants of electoral outcomes’, Annual

14

Review of Political Science 3: 183–219. Maloney, Patrick. 2010 “This is about moving forward” article from The London Free Press. October 27,

2010. http://www.lfpress.com/news/london/2010/10/27/15850251.html Meisel, J. 1973. Working Papers on Canadian Politics. Enlarged ed. Montreal: McGill-Queen’s University

Press. Mendelsohn, M. and J. S. Matthews. 2010. “The New Ontario: The Shifting Attitudes of Ontarians

toward the Federation.” The Mowat Centre for Policy Innovation. Miner, John. 2010 “Fontana vows 10,000 jobs within five years” article from The London Free Press.

Ocober 6, 2010. http://www.lfpress.com/news/london/2010/10/08/15627556.html Nadeau, R. and A. Blais. 1993. “Explaining Election Outcomes in Canada: Economy and Politics.”

Canadian Journal of Political Science 26: 775-790. Nadeau, R. and A. Blais.1995. “Economic Conditions, Leader Evaluations and Election Outcomes in

Canada.” Canadian Public Policy 21 (2): 212-218. Nannestad, P. and M. Paldam. 1994. “The VP Function: A Survey of the Literature on Vote and Popularity

Functions after 25 Years.” Public Choice 79: 213-45. Nevitte, N., A. Blais, E. Gidengil and R. Nadeau. 2000. Unsteady State: The 1997 Canadian Federal

Election. Don Mills: Oxford University Press. Norpoth, H. 1992. Confidence Regained: Economics, Mrs. Thatcher and the British Voter. University of

Michigan Press: Ann Arbor. Powell, G. B. and G. Whitten. 1993. “A Cross-National Analysis of Economic Voting: Taking Account of

Political Context.” American Journal of Political Science 37 (2): 391-414. Sancton, A. 2011. Canadian Local Government. Toronto: Oxford University Press. Stanwick, H. 2000. “A Megamayor for All People? Voting Behaviour and Electoral Success in the 1997

Toronto Municipal Election.” Canadian Journal of Political Science 33, 3: 549-68. Stokes, D. E. 1963. “Spatial Models of Party Competition.” American Political Science Review 57, 2: 368-

77. Stokes, D. E. 1992. “Valence Politics.” In Electoral Politics, ed. Dennis Kavanagh. Oxford: Clarendon Press. Tellier, G. 2006. “Effect of Economic Conditions on Government Popularity: The Canadian Provincial

Case.” Canadian Journal of Political Science 39 (1): 27-51. Vermeir, J. and B. Heyndels. 2006. “Tax Policy and Yardstick Voting in Flemish Municipal Elections.”

Applied Economics 38, 19: 2285-98.

15

Table 1 Summary Statistics of Variables

Dependent Variable Mean Standard Deviation Maximum Minimum

Vote Percentage 53.02 21.99 100 3.58

Independent Variables

Unemployment Rate 6.89 1.95 14.19 3.22

Terms in Office 1.64 0.87 4 1

Candidates Running 6.01 5.06 38 0

Ontario - - 47 (Ont.) 43 (not Ont.)

Past vote 56.71 16.41 100 28

Table 2 Incumbent Mayoral Vote Percentage

Model 1 Model 2 Model 3

Unemployment -1.62 (1.18)‡ -1.30 (1.19) -0.75 (1.35)

Terms -3.26 (2.68) -3.60 (2.66) -4.09 (3.15)‡

Candidates -0.88 (0.46)* -0.74 (0.46)‡ -0.51 (0.47)

Ontario - -7.78 (4.68)‡ -10.72 (5.15)**

Past Vote - - 0.13 (.16)

Constant 74.83 (10.63)*** 76.41 (10.56)*** 76.41 (10.56)***

R2 0.07 0.09 0.09

N 90 90 77 Note: Cells contain coefficients from OLS regression and standard errors in parentheses.

p<.2‡ p<.1* p<.05** p<.01***

Table 3 Incumbent Mayoral Vote Percentage: Inside and Outside of Ontario

Ontario Outside Ontario

Unemployment 1.03 (1.93) 0.65 (2.14) -4.29 (1.39)*** -3.07 (1.71)*

Terms -2.13 (3.75) -0.31 (4.68) -5.70 (3.38)‡ -5.12 (4.15)

Candidates -0.22 (0.49) -0.08 (0.53) -4.74 (1.21)*** -3.99 (1.32)***

Quebec - - 0.19 (9.14) -1.75 (9.53)

Past Vote - -0.17 (.27) - 0.15 (0.20)

Constant 46.05 (16.32)*** 52.16 (21.34)** 119.26 (14.39)*** 98.05 (20.14)***

R2 0.02 0.02 0.37 0.32

N 47 39 43 38 Note: Cells contain coefficients from OLS regression and standard errors in parentheses.

p<.2‡ p<.1* p<.05** p<.01***

16

Appendix 112 List of Census Metropolitan Areas (CMA) and election year included in the analyses of this paper: St. John’s (1997, 2001, 2005, 2009); Halifax (2004, 2008); Moncton (1998, 2001); Saint John (1998, 2001, 2004, 2008); Montreal (2009); Quebec City (2009); Saguenay (2009); Trois-Rivieres (2009); Gatineau (2009); Ottawa (2003, 2006, 2010); Kingston (2000, 2003, 2006); Peterborough (2000, 2003, 2010); Oshawa (1997, 2000, 2003, 2006, 2010); Toronto (2000, 2006); Hamilton (2006, 2010); St. Catharine's (2000, 2003, 2010); Brantford (1997, 2000, 2003, 2006); Guelph (1997, 2003, 2006, 2010); London (1997, 2003, 2006, 2010); Windsor (1997, 2003, 2006, 2010); Barrie (1997, 2000, 2003, 2006, 2010); Sudbury (2006, 2010); Thunder Bay (2000, 2006, 2010); Winnipeg (2002, 2006, 2010); Saskatoon (2000, 2003, 2006, 2009); Regina (1997, 2000, 2003, 2006, 2009); Edmonton (1998, 2001, 2004, 2007, 2010); Calgary (2004, 2007); Kelowna (1999, 2002, 2005, 2008); Vancouver (1999); Victoria (2002, 2005)

Appendix 2 List of sources for municipal election data: Outside Ontario St. John’s: http://www.stjohns.ca/cityhall/elections/index.jsp Halifax: http://www.halifax.ca/election/ElectionArchives/index.html Moncton and Saint John: http://www.gnb.ca/elections/publications-e.asp#3 Montreal, Quebec City, Saguenay, Trois-Rivieres and Gatineau: http://www.electionsmunicipales.gouv.qc.ca/resultats/resu/afficher_resultats.php Winnipeg: http://winnipeg.ca/clerks/docs/election_services/election_archive.stm Saskatoon: http://www.saskatoon.ca/CITY%20COUNCIL/Elections/Pages/PastElections.aspx Regina: Received through information request submitted at City of Regina website. Edmonton: http://www.edmonton.ca/city_government/municipal_elections/election-history.aspx Calgary: Received through information request submitted at City of Calgary website. Kelowna: http://www.city.kelowna.bc.ca/CM/Page161.aspx Vancouver: http://data.vancouver.ca/datacatalogue/municipalElectionResults.htm Victoria: http://www.victoria.ca/cityhall/departments_crpleg_abt.shtml

Ontario Ottawa: http://ottawa.ca/city_hall/elections/ Kingston: http://www.cityofkingston.ca/cityhall/election/index.asp and Direct Contact with City Clerk’s Office Peterborough: Direct Contact with City Clerk’s Office Oshawa: http://www.oshawa.ca/election/2010/prior.asp and Direct Contact with City Clerk’s Office Toronto: http://www.toronto.ca/elections/ Hamilton: http://www.hamilton.ca/CityDepartments/CorporateServices/Clerks/MunicipalElection/ and Direct Contact with City Clerk’s Office St. Catharines-Niagara: http://www.stcatharines.ca/en/governin/Elections.asp?_mid_=9971

12

Please note that Kitchener, Cambridge and Waterloo were not included in the data set. The CMA data available for these areas of Ontario provides one measure of unemployment despite the very different economic situations of each municipal entity.

17

Brantford: http://www.brantford.ca/govt/elections/Pages/ElectionResults.aspx Guelph: http://www.guelph.ca/cityhall.cfm?SubCatId=2232&smocid=2805 and http://guelph.ca/council/index.php/?page_id=408 London: http://london.ca/d.aspx?s=/Elections/election_info.htm and Direct Contact with City Clerk’s Office Windsor: http://www.citywindsor.ca/003241.asp Barrie: http://www.barrie.ca/City%20Hall/election/Pages/ElectionInformation.aspx and Direct Contact with City Clerk’s Office Sudbury: http://elections.greatersudbury.ca Thunder Bay: http://www.thunderbay.ca/City_Government/Thunder_Bay_Votes_2010.htm and Direct Contact with City Clerk’s Office